A place for discussion about anything related to the Recogito annotation platform
March 29, 2016 at 9:42 am #619
[This could be cross posted in the Gazetteer forum]
Leif wrote to Humphrey ‘pastplace‘ Southall recently, about corrections or additions to his gazetteer. Has he replied yet?
I wonder, in any case, whether Recogito might benefit from its own gazetteer-in-limbo. Places that can’t be found in the main gazetteers could be logged (name and coordinates), and remain until someone sees fit to add them to an authorised gazetteer. At least we’d have a position within Recogito, even if it isn’t yet linked to the outside world.
September 21, 2016 at 1:22 am #1939
I have a related query: is it possible to add other search possibilities (other gazetteer results) to Recogito for those of us who aren’t working on the ancient world? Or, would it be best to adapt Recogito code outside the main project?
I’m in the process of considering different methods for annotating 17th and 18th c. texts that would be the foundation for an early modern French gazetteer. As I do this, I’d like to be able to draw on existing data so I don’t duplicate what’s already out there (http://histograph.io/ for the Netherlands, for example, will have some overlap). Recogito seems like a user-friendly platform for this, though we’re also exploring other options.
September 21, 2016 at 11:50 am #1940
- Our own ‘official’ server at http://recogito.pelagios.org will get GeoNames as a third gazetteer option (in addition to the current Pleiades & Digital Atlas of the Roman Empire) very soon. Hopefully, this will happen next week. I’m still working on filtering the data down to a more compact subset of GeoNames, though.
- If you are happy to set up your own copy of Recogito: you can ingest your own set of gazetteers of course. You’ll need to have (or prepare) a dump file of each gazetteer in a specific format, though. There’s an RDF-based one (originally defined by Pelagios); and there’s also the new GeoJSON-based format of the Pleiades gazetteer. Both are, more or less, supported by Recogito (Pleiades requires a small conversion step to make it “streamable” into Recogito). So if you have the possibility to get your gazetteer into either one, you’re good to go. (If, for whatever reason, this is strictly not an option: adding a custom crosswalk to the Recogito code wouldn’t be such a big deal either. But in the spirit of not letting things fragment too much, I’d love to see more gazetteers being published in compatible formats.)
- Last but not least, there’s also an ongoing discussion of enabling people to upload their own data tables of places (as a document in Recogito) and then creating cross references to that table in the texts. Not quite gazetteer tagging; but possibly an option for smaller-sized & highly specialized datasets of places. That’s still very much in the idea/conceptual stage though.
September 23, 2016 at 1:18 am #1949
Hi Rainer – Thanks very much for the detailed reply. This is really helpful. I have a few questions/thoughts to follow up.
- When you say that you are filtering down Geonames, what is your method? We’re currently working on matching Geonames records to 18th c. place names, but I imagine you’re not using quite so limited a filter.
- Talking with my colleagues yesterday, it seems like we might want to adapt a local version of Recogito to hand annotate certain sources. For the moment, I’m most interested in the DEEP (English Place Names), Geonames, and other European gazetteers (Netherlands, Germany, Switzerland). I will have to check about what format we can access them in. We are considering different methods for creating a gold standard set of training data for early modern French to assist with NER with the Edinburgh Geoparser.
- Regarding your last point (if I understand correctly), I think we will be dealing with too many place names to make that kind of solution feasible. On the other hand, incorporating more gazetteers that accommodate other geographical areas/chronological periods would be great. One issue with our early modern documents, for example, is that I’d like to be able to tag place names that are temporally specific. E.g. in the Diderot/d’Alembert Encyclopédie article “Paris”, there are mentions both of Paris and of Lutèce. It would be nice to tag “Paris” as 18th c. and “Lutèce” as Gallo-Roman.
September 23, 2016 at 7:30 am #1951
- with regard to GeoNames: the main goal for now is really to have a compact, general-purpose dataset to start with. I’m using GeoNames’ own cities dataset (all cities with population > 1.000), merging that with the “alternateNames” table (also a separate GeoNames dataset). Furthermore, we’re merging in records and boundary outlines for countries (two more separate GeoNames datesets). I’m not too happy with the boundaries yet, so will still be looking for a better dataset there; and then I’m also looking for a good source for at least a first level (states etc.) administrative hierarchy below countries. That, more or less, I think would work as a basic “get-started” dataset for our first Recogito release in December.
- Do let us know about your progress with those datasets. As you can imagine, I’m highly interested in re-using them!
- Yes, I had the impression the data-table approach would be insufficient in your case. Regarding tagging different time periods. I think there’s two ways to approach it. One – as you say – is to use different gazetteers, each with its own specific temporal/cultural scope. Another way could be to tag with the same gazetteer, but use plain-old keyword tagging alongside to make the distinction. (Keyword tagging is one of the upcoming features. I.e. just like adding a freetext comment, you could add an “18thCentury” or a “GalloRoman” tag or similar.)
September 26, 2016 at 6:54 am #1956
Our Geonames database is everything from the all countries file + alternate names. It will be interesting to compare our results vs. yours with the more limited cities dataset. Working with Geonames is a bit of a headache, but we’re also lucky to have it. You take the good with the bad!
That’s good news about the tagging feature – I will make use of it with more annotation tests. Looking forward to the new Recogito release.
Any chance you or your colleagues are going to be at the Creating Spatial Historical Knowledge conference in DC next month? I’ll be presenting on our 18th c. France work.
September 26, 2016 at 8:23 am #1957
Working with Geonames is a bit of a headache, but we’re also lucky to have it.
I couldn’t phrase it any better
Meanwhile, I added what we have to our Recogito instance at http://recogito.pelagios.org, so GeoNames tagging with the limited dataset should now work.
October 3, 2016 at 6:53 am #1973
I don’t know about the conference in DC. @kmcdono2 – can you let us know what it is, and then perhaps we can find a suitable SIG form on which to pass on the information? If no one else from the community can go, you are also very welcome to mention the project in your presentation (if it’s useful for you to do so). We have (many!) slides we can share…
October 3, 2016 at 7:08 am #1976
Hi Elton – The conference is at the German Historical Institute co-sponsored by Berkeley DH and CHNM. I just got a draft program last week so hopefully they will post one on the GHI site soon.
If there is a slide about the latest developments for Recogito, or a summary of what you hope it will be doing in the future to include gazetteers from other periods, that would be fantastic. I’m hoping to show that we early modernists can learn from and support the work ancient/medieval scholars have done with digital gazetteers.
I went in to try Recogito now that the Geonames gazetteer is an option. I was able to add in many of the more “modern” place names that appear in some of my 18th c. texts. Immensely useful!
October 3, 2016 at 7:45 am #1980
October 4, 2016 at 11:42 pm #1981
October 5, 2016 at 6:39 pm #1986
Great. @rainer has the latest presentations: he’ll send you a link to his slideshare. If you were happy to, it might be an idea to post a brief sketch of your research plans either here or, I guess, in the Modern SIG. That way people who might actually know something about the period, as well as those who more informed about the technical side of things, could contribute. What do you think?
October 6, 2016 at 12:16 am #1987
Sounds great. I posted an initial query over on the Gazetteers SIG a few months ago. @kgeographer and I have been talking a bit about the technical side of things with the early modern world gazetteer. I’d like to post an update soon with some of the work I’ve been doing with the digital Encyclopédie. Could do so in the Modern SIG.
October 6, 2016 at 6:43 am #1991
I put the slides for a presentation I gave recently here:
It’s mostly a general overview of Pelagios. But with a bit of an extra gazetteer focus, so hopefully this is something you can use and/or pick from. Beware: it’s a big file (44MB) because there are two videos embedded: one of our prototype search engine Peripleo (work on this will start again next year), and one from Recogito (in the state it was in July).
Let me know if that’s useful and/or if you need anything else!
P.S.: there’s also a PDF version available via http://gis.harvard.edu/events/seminar-series/seminar-linked-places-and-modeling-historical-movement. This has only the slides, and links to online versions of the screencasts on vimeo.
- This reply was modified 2 years, 6 months ago by Rainer Simon.
You must be logged in to reply to this topic.